Polymerase chain reaction (PCR) is a technique widely used in molecular biology. It derives its name from one of its key components, a DNA polymerase used to amplify a piece of DNA by in vitro enzymatic replication. As PCR progresses, the DNA generated (the amplicon) is itself used as a template for replication. This sets in motion a chain reaction in which the DNA template is exponentially amplified. With PCR, it is possible to amplify a single or few copies of a piece of DNA across several orders of magnitude, generating millions or more copies of the DNA piece. PCR employs a thermostable polymerase, dNTPs, and a pair of primers.
PCR is conceptually divided into 3 reactions, each usually assumed to occur over time at each of three temperatures. Such an “equilibrium paradigm” of PCR is easy to understand in terms of three reactions (denaturation, annealing, and extension) occurring at 3 temperatures over 3 time periods each cycle. However, this equilibrium paradigm does not fit well with physical reality. Instantaneous temperature changes do not occur; it takes time to change the sample temperature. Furthermore, individual reaction rates vary with temperature, and once primer annealing occurs, polymerase extension immediately follows. More accurate, particularly for rapid PCR, is a kinetic paradigm where reaction rates and temperature are always changing. Holding the temperature constant during PCR is not necessary as long as the products denature and the primers anneal. Under the kinetic paradigm of PCR, product denaturation, primer annealing, and polymerase extension may temporally overlap and their rates continuously vary with temperature. Under the equilibrium paradigm, a cycle is defined by 3 temperatures each held for a time period, whereas the kinetic paradigm requires transition rates and target temperatures. Illustrative time/temperature profiles for the equilibrium and kinetic paradigms are shown in FIGS. 15a-15b. However, it is understood that these temperature profiles are illustrative only and that in some implementations of PCR, the annealing and extension steps are combined so that only 2 temperatures are needed.
Paradigms are not right or wrong, but they vary in their usefulness. The equilibrium paradigm is simple to understand and lends itself well to the engineering mindset and instrument manufacture. The kinetic paradigm is more relevant to biochemistry, rapid cycle PCR, and melting curve analysis.
When PCR was first popularized in the late 1980s, the process was slow. A typical protocol was 1 minute for denaturation at 94° C., 2 minutes for annealing at 55° C., and 3 minutes for extension at 72° C. When the time for transition between temperatures was included, 8 minute cycles were typical, resulting in completion of 30 cycles in 4 hours. Twenty-five percent of the cycling time was spent in temperature transitions. As cycling speeds increased, the proportion of time spent in temperature transitions also increased and the kinetic paradigm became more and more relevant. During rapid cycle PCR, the temperature is usually changing. For rapid cycle PCR of short products (<100 bps), 100% of the time may be spent in temperature transition and no holding times are necessary. For rapid cycle PCR of longer products, a temperature hold at an optimal extension temperature may be included.
In isolation, the term “rapid PCR” is both relative and vague. A 1 hour PCR is rapid compared to 4 hours, but slow compared to 15 minutes. Furthermore, PCR protocols can be made shorter if one starts with higher template concentrations or uses fewer cycles. A more specific measure is the time required for each cycle. Thus, “rapid cycle PCR” (or “rapid cycling”) was defined in 1994 as 30 cycles completed in 10-30 minutes (1), resulting in cycles of 20-60 seconds each. This actual time of each cycle is longer than the sum of the times often programmed for denaturation, annealing and extension, as time is needed to ramp the temperatures between each of these stages. Initial work in the early 1990s established the feasibility of rapid cycling using capillary tubes and hot air for temperature control. Over the years, systems have become faster, and the kinetic requirements of denaturation, annealing, and extension have become clearer.
In one early rapid system, a heating element and fan from a hair dryer, a thermocouple, and PCR samples in capillary tubes were enclosed in a chamber (2). The fan created a rapid flow of heated air past the thermocouple and capillaries. By matching the thermal response of the thermocouple to the sample, the temperature of the thermocouple closely tracked the temperature of the samples, even during temperature changes. Although air has a low thermal conductivity, rapidly moving air against the large surface area exposed by the capillaries was adequate to cycle the sample between denaturation, annealing, and extension temperatures. Electronic controllers monitored the temperature, adjusted the power to the heating element, and provided the required timing and number of cycles. For cooling, the controller activated a solenoid that opened a portal to outside air, introducing cooling air to the otherwise closed chamber.
Temperatures could be rapidly changed using the capillary/air system. Using a low thermal mass chamber, circulating air, and samples in glass capillaries, PCR products >500 bp were visualized on ethidium bromide stained gels after only 10 minutes of PCR (30 cycles of 20 seconds each) (3). Product yield was affected by the extension time and the concentration of polymerase. With 30 second cycle times (about 10 seconds between 70 and 80° C. for extension), the band intensity increased as the polymerase concentration was increased from 0.1 to 0.8 Units per 10 μl reaction. It is noted that polymerase Unit definitions can be confusing. For native Taq polymerase, 0.4 U/10 μl is about 1.5 nM under typical rapid cycling conditions (50).
Rapid protocols use momentary or “0” second holds at the denaturation and annealing temperatures. That is, the temperature-time profiles show temperature spikes for denaturation and annealing, without holding the top and bottom temperatures. Denaturation and annealing can occur very quickly.
Rapid and accurate control of temperature allowed analytical study of the required temperatures and times for PCR. For an illustrative 536 bp fragment of human genomic DNA (β-globin), denaturation temperatures between 91° C. and 97° C. were equally effective, as were denaturation times from <1 second to 16 seconds. However, it was found that denaturation times longer than 16 seconds actually decreased product yield. Specific products in good yield were obtained with annealing temperatures of 50-60° C., as long as the time for primer annealing was limited. That is, best specificity was obtained by rapid cooling from denaturation to annealing and an annealing time of <1 second. Yield was best at extension temperatures of 75-79° C., and increased with extension time up to about 40 seconds.
Conclusions from this early work were: 1) denaturation of PCR products is very rapid with no need to hold the denaturation temperature, 2) annealing of primers can occur very quickly and annealing temperature holds may not be necessary, and 3) the required extension time depends on PCR product length and polymerase concentration. Also, rapid cycle PCR is not only faster, but better in terms of specificity and yield (4, 5) as long as the temperature was controlled precisely. PCR speed is not limited by the available biochemistry, but by instrumentation that does not control the sample temperature closely or rapidly.
However, most current laboratory PCR instruments perform poorly with momentary denaturation and annealing times, and many don't even allow programming of “0” second holding periods. Time delays from thermal transfer through the walls of conical tubes, low surface area-to-volume ratios, and heating of large samples force most instruments to rely on extended times at denaturation and annealing to assure that the sample reaches the desired temperatures. With these time delays, the exact temperature vs time course becomes indefinite. The result is limited reproducibility within and high variability between commercial products (6). Many instruments show marked temperature variance during temperature transitions (7, 8). Undershoot and/or overshoot of temperature is a chronic problem that is seldom solved by attempted software prediction that depends on sample volume. Such difficulties are compounded by thermal properties of the instrument that may change with age.
Over time, conventional heat block instruments have become faster, with incremental improvements in “thin wall” tubes, more conductive heat distribution between samples, low thermal mass blocks and other “fast” modifications. Nevertheless, it is unusual for these systems to cycle rapidly enough to complete a cycle in less than 60 seconds. A few heat block systems can achieve <60 second cycles, usually restricted to 2-temperature cycling between a limited range of temperatures. By flattening the sample container, rapid cycling can be achieved by resistive heating and air cooling (9), or by moving the sample in a flexible tube between heating zones kept at a constant temperature (U.S. Pat. No. 6,706,617).
Commercial versions of the air/capillary system for PCR have been available since 1991 (1) and for real-time PCR since 1996 (10μ, 11). Rapid cycling capabilities of other instruments are often compared against the air/capillary standard that first demonstrated 20-60 second cycles. Oddly enough, there has been a trend to run the capillary/air systems slower over the years, perhaps reflecting discomfort with “0” second denaturation and annealing times by many users. Also, heat-activated enzymes require long activation periods, often doubling run times even when “fast” activation enzymes are used. Another compromise away from rapid cycling is the use of plastic capillaries. These capillaries are not thermally matched to the instrument, so 20 second holds at denaturation and annealing are often required to reach the target temperatures (12).
Some progress in further decreasing the cycle times for PCR has occurred in microsystems, where small volumes are naturally processed (13, 14). However, even with high surface area-to-volume sample chambers, cycles may be long if the heating element has a high thermal mass and is external to the chamber (15). With thin film resistive heaters and temperature sensors close to the samples, 10-30 minute amplification can be achieved (16, 17).
While cooling of low thermal mass systems is usually by passive thermal diffusion and/or by forced air, several interesting heating methods have been developed. Infrared radiation can be used for heating (18) with calibrated infrared pyrometry for temperature monitoring (19). Alternatively, thin metal films on glass capillaries can serve as both a resistive heating element and a temperature sensor for rapid cycling (20). Finally, direct Joule heating and temperature monitoring of the PCR solution by electrolytic resistance is possible and has been implemented in capillaries (21). All of the above methods transfer heat to and from fixed samples.
Instead of heat transfer to and from stationary samples, the samples can be physically moved to different temperature baths, or through channels with fixed temperature zones. Microfluidic methods have become popular, with the PCR fluid passing within channels through different segments kept at denaturation, annealing, and extension temperatures. Continuous flow PCR has been demonstrated within serpentine channels that pass back and forth through 3 temperature zones (22) and within loops of increasing or decreasing radius that pass through 3 temperature sectors (23). A variant with a serpentine layout uses stationary thermal gradients instead of isothermal zones, to more closely fit the kinetic paradigm of PCR (24). To limit the length of the microchannel necessary for PCR, some systems shuttle samples back and forth between temperature zones by bi-directional pressure-driven flow (25), pneumatics (26), or electrokinetic forces (27). Instead of linear shuttling of samples, a single circular channel can be used with sample movement driven as a magnetic ferrofluid (28) or by convection (29). One potential advantage of microsystem PCR, including continuous flow methods, is cycling speed.
Although some microsystems still require >60 second cycles, many operate in the 20-60 second cycle range of rapid cycle PCR (13, 30). Minimum cycle times ranging from 16-37 seconds have been reported for infrared heating (18, 19). Metal coated capillaries have achieved 40 second PCR cycles (20), while direct electrolytic heating has amplified with 21 second cycles (20). Minimum cycle times reported for closed loop convective PCR range from 24-42 seconds (29, 31). Several groups have focused on reducing PCR cycle times to <20 seconds, faster than the original definition of rapid cycle PCR that was first demonstrated in 1990. Thin film resistive heating of stationary samples has reduced cycle times down to 17 seconds for 25 μl samples (32) and 8.5 seconds for 100 nl samples (17). Continuous flow systems have achieved 12-14 second cycles with thermal gradient PCR (24) and sample shuttling (26), while a ferrofluid loop claims successful PCR with 9 second cycles (28). Continuous flow systems through glass and plastic substrates have achieved cycle times of 6.9 seconds (22) and 5.2 seconds (23) for various size PCR products. Alternating hot and cool water conduction through an aluminum substrate amplified 1 μl droplets under oil with 5.25 second cycles (33). Similarly, water conduction through a porous copper block amplified 5 μl samples with 4.6 second cycles (34). A continuous flow device of 1 μl reaction plugs augmented by vapor pressure achieved 3 second cycles (35). Additionally, there are reports that claim to amplify an 85 bp fragment of the Stx bacteriophage of E. coli in capillaries with 2.7 second cycles by immersion of the capillaries in gallium sandwiched between Peltier elements (36). Alternatively, PCR amplification in capillaries cycled by pressurized hot and cool gases obtained 2.6 second cycles (48).
Table 1 summarizes work to minimize PCR cycle times to less than the 20 second cycles that originally defined “Rapid PCR”. Over the past 20 years, new prototype instruments have been developed that incrementally improve cycling speed. However, practical PCR performance (efficiency and yield) is often poor. As a general rule, as cycles become increasingly shorter, claims for successful PCR correlate with lower complexity targets (bacteria, phage, multicopy plasmids, or even PCR products) that are used at higher starting concentrations (see, e.g., U.S. Pat. No. 6,210,882, wherein 5 ng of amplicon was used as the starting sample). Indeed, none of the studies listed in Table 1 with <20 second cycles used complex eukaryotic DNA such as human DNA. The starting copy number of template molecules is often very high (e.g., 180,000,000 copies of lambda phage/μl), so that little amplification is needed before success is claimed. Furthermore, the lack of no template controls in many studies raises questions regarding the validity of positive results, especially in an environment with high template concentrations. One instrument-oriented report focuses extensively on the design and modeling of the thermal cycling device, with a final brief PCR demonstration using a high concentration of a low complexity target. Heating and cooling rates (up to 175° C./s) have been reported based on modeling and measurements without PCR samples present (17).
TABLE 1FastestCycleTotalProductTime[Template][Primers][Polymerase]Length(s)(Copies/μl)Template Form(nM)Polymerase(nM)(bp)20 1,600Human DNA10000.08 U/μl Taq 35361240,000Lambda phage4000.2 U/μl Taq  7.5500121,000,000  230 bp PCR10000.5 U/μl Taq19230product9.254,700-18S rDNA1800Taq Gold?187470,000(humangenomic)918,000,000  Lambda phage20000.025 U/μl Taq   0.945008.5?cDNA1800??827.010,000,000  1 KB PCR20000.25 U/μl Taq  9.4176product6.310,000Plasmids (B.12000.05 U/μl Ex?134anthracis)Taq HS5.2/9.7180,000,000   Lambda phage4000.07 U/μl Taq  2.6500/9975.251,400,000  B. subtilis5000.025 U/μl?72(bacterialKOD plusDNA)4.634,000E. herbicola8000.04 U/μl 4 58/160(bacterialKAPA2GDNA)4.2   501B. subtilis? KOD plus?72(bacterialDNA)3.010,000Plasmids (B.12000.05 U/μl Ex?134anthracis)Taq HS2.7?stx phage (E.?3KOD?85coli)2.6 ?4stx phage (E.?50.5 U/μl Taq1985coli)FastestCycleNoTimeTemplate(s)QuantificationTrendMethodControl?Reference20Faint Gel BandIncreases withCapillary Air CyclingNo3[Polymerase]12Capillary?IR Heating, PressurizedNo56ElectrophoresisAir Cooling12Good gel bandDependent on cycle #Continuous FlowYes55and copy #9.25??IR Heating of droplets inNo54oil9OK gel bandIntensity increasesContinuous Flow with aNo28with cycle timeFerrous Particle Plug8.580% efficiencyDecreasing efficiencyMicromachined?17at faster cyclescantilever7.07% of control50% at 15 s cyclesContinuous FlowYes226.355% of control?Plug Continuous FlowYes535.2/9.7Faint gel bandsDependent on cycleContinuous FlowNo23times5.2590% efficiencySingle runWater pumped againstYes33(SYBR)aluminum plate4.6Faint gel bandsYield increases with #Water pumped throughNo31cyclesporous copper4.2Cq = 33 (SYBR)Higher copy # reducesIR laser?251Cq3.015% of control80% at 7.5 s cyclesConstant flow with vaporYes (5%35pressuresignal)2.7Barely visible bandDecreasing yield fromGallium transfer fromNo363.06 s to 2.69 s cyclesPeltiers to capillaries2.6Very dim bandConstant from 2.8 toPressurized gas andNo482.6 s cyclescapillaries1Presumed single copy in a 20 nl droplet with Cq of 33 under SYBR Green monitoring, but no gel or melting analysis to confirm PCR product identity.2A “Blank” sample was run, but it is not clear if this was a no template control.3Article says [primer] is 0.5 mmol, patent application (US 2009/0275014 A1) says [primer] is 0.01-0.5 μM.4Two pg E. coli DNA/μl, but copy number of phage in the DNA preparation is unknown.5Dissertation says 0.5 μmol/10 μl (50 mM), patent (U.S. Pat. No. 6,472,186) says 50 pmol/10 μl (5 μM).
One way to decrease cycle time is to introduce variations to the PCR protocol to ease the temperature cycling requirements. Longer primers with higher Tms allow higher annealing temperatures. By limiting the product length and its Tm, denaturation temperatures can be lowered to just above the product Tm. In combination, higher annealing and lower denaturation temperatures decrease the temperature range required for successful amplification. Reducing 3-step cycling (denaturation, annealing, and extension) to 2-steps (denaturation and a combined annealing/extension step) also simplifies the temperature cycling requirements. Both decreased temperature range and 2-step cycling are typical for the studies in Table 1 with cycle times <20 seconds. Two-step cycling can, however, compromise polymerase extension rates if the combined annealing/extension step is performed at temperatures lower than the 70 to 80° C. temperature optimum where the polymerase is most active. Polymerase extension rates are log-linear with temperature until about 70-80° C., with a reported maximum of 60-120 bp/s (50).
Even with protocol variations, amplification efficiency and yield are often poor when cycle times are <20 seconds when compared to control reactions (22, 23). These efforts towards faster PCR appear dominated by engineering with little focus on the biochemistry. As cycle times decrease from 20 seconds towards 2 seconds, PCR yield decreases and finally disappears, reflecting a lack of robustness even with simple targets at high copy number.
The instrumentation in various references disclosed in Table 1 may be suitable for extremely fast PCR, if reaction conditions are compatible. As disclosed herein, a focus on increased concentrations of primers, polymerase, and Mg++ allows for “extreme PCR” (PCR with <20 second cycles (30 cycles in <10 min)), while retaining reaction robustness and yield.